Version HTML de base
18
“Uncertainty-noise” Le Mans
Acoustique
&
Techniques n° 40
Treatment of Measurement Uncertainties in International and European Standards on Acoustics
major uncertainty contributions and will enable the laboratory
to improve certain measurement conditions in order to reduce
the overall uncertainty, if requested.
6- If it appears during the preparation of a standard that
sufficient knowledge is not yet available to fully apply the
GUM while solid information on reproducibility data from
interlaboratory comparisons is available, this information shall
be given. A statement shall then be included in the main text
that the expanded uncertainty for a stated coverage probability
of 95 % shall be given as two times the standard deviation of
reproducibility.
NB : If a coverage probability different from the preferred 95
% is chosen, this factor has to be adjusted accordingly.
This approach is, however, to be regarded as intermediate
only and shall be replaced by the expression of uncertainty in
conformity with the GUM in any future revision of the standard.
Guidance on how the GUM may be principally applied, as
described under item 4 above, shall be included in the standard
already now.”
Additional interpretation
Paragraph 1:
Each standard specifying a method for the measurement or
prediction of sound, be it new or a revision of an existing
standard, shall contain a clause on measurement uncertainty
as given. It requests that the uncertainty has to be evaluated
when the standardized method is applied. Preference is give
to the GUM. However, in recognition of the fact that present
knowledge might not always allow application of the GUM
quantitatively, other methods of evaluation are permitted (see
paragraph 6 below). It is further recognized that, in accordance
with ISO/IEC 17025, the measurement uncertainty need not
always be stated in test reports. However, if reported, some
uniformity is needed. Therefore, the expanded measurement
uncertainty together with the chosen coverage probability
(preferably 95%) as specified in the GUM shall be stated.
Moreover, each standard shall contain an annex where the
concept of the GUM is described in detail for the given
application allowing a laboratory to apply these evaluation
principles in cases where definite figures on the various
uncertainty sources cannot be specified at the present
stage.
Paragraph 2:
The text underlines the need to refer to the GUM when applying
the given standard.
Paragraph 3:
See the interpretation remarks on paragraph 1.
Paragraph 4:
As further guidance to Working Groups, an outline of the
contents of the uncertainty annex is given. It especially
mentions the need to formulate a functional relationship
describing the measurements and to establish an uncertainty
budget preferably in tabular form.
A recently published standard developed in TC 43/SC 1, i.e.
ISO 3745:2003, Acoustics – Determination of sound power
levels of noise sources using sound pressure – Precision
method for anechoic and hemi-anechoic rooms, is mentioned
as providing a valuable example on how to structure and draft
the uncertainty annex. This example is considered to be easily
adaptable to other measurement standards in acoustics.
Paragraph 5:
It is emphasized that it is finally the responsibility of a testing
laboratory to evaluate its uncertainty even if a standardized
method is applied. There may always be certain deviations
from the given method and, on the other hand, specific
measurement conditions in a laboratory might be more
favorable than generally allowed. The Working Group may
decide whether it is appropriate to include figures in the
uncertainty budget and whether, in the affirmative, such figures
are to be considered as worst case or typical.
In any case, the attempt to evaluate the relevant sources of
uncertainty according to the GUM is of great benefit for the
laboratory since it allows detection of the main contributions
to the overall uncertainty and may then lead to a reduction of
the uncertainty if requested by the customer of the test.
Paragraph 6:
It is recognized that present knowledge may not always
be sufficient to apply the GUM in each quantitative detail
(see paragraph 1 above). On the other hand, reliable data
from interlaboratory comparisons are sometimes available
providing valid data on reproducibility which occurs if the
method is applied by different laboratories. The adopted
policy, therefore, allows proper use to be made of these data
for statements of the measurement uncertainty in test reports
as a pragmatic and intermediate step in order not to hinder the
ongoing standardization process. However, the prerequisite for
using such data is that solid data from intercomparisons using
the specified method really exist and are not just estimated.
Furthermore, it has always to be realized that measurement
uncertainty based on reproducibility data may provide a too
optimistic picture since not all uncertainty sources may have
been become apparent in the interlaboratory comparisons.
Recent experience with the implementation of
the TC 43 policy paper
Experience in a Working Group responsible for basic
measurement standards on machinery noise
In a current revision of all the standards in the series ISO 3740-
ISO 3747, Acoustics – Determination of sound power levels
and sound energy levels of a noise source using measurements
of sound pressure – the practice already adopted in ISO
3745:2003 mentioned above has been further developed.
Most of the revised standards in the series are at present
at the stage of 2nd Committee Draft. In this form, they each
include a mandatory clause on measurement uncertainty,
which states that:
1- Measurement uncertainties shall normally be evaluated in
conformity with the GUM. However, the information necessary
to achieve this is not at present available and guidance is given
in an informative annex.
2- Standard deviations of reproducibility, relevant to each
individual standard in the series, are given based on published
laboratory intercomparisons.